Course Wiki
Time Series Decomposition
Splitting a price series into trend, seasonality, and remainder — the SMA modeling module's first lens.
Related Pages
Module 3 opens modeling by decomposing a price series into additive parts: the long-run trend, repeating seasonal patterns, and the remainder — the residual movement left after the predictable components are removed. The decomposition makes the series’ structure visible and frames what a model can realistically predict.
From there the module layers prediction approaches: heuristic and hand rules first, regression for relationships, and binary classification to predict the direction of growth. An optional neural-network example closes the module. The constant is the framing — predict the future growth window, and evaluate honestly against it.
Taught in
- Stock Markets Analytics Zoomcamp — Module 3: Modeling for Time Series